ES2588745T3 - Encoding method, decoding method, encoder device, decoder device, program and recording medium - Google Patents

Abstract

A coding method comprising: a vector quantification stage for the collective vector quantification of several samples to obtain a vector quantification index and the quantized value of each of the several samples, and a stage for selecting groups of production coefficients Index Information indicating a group of coefficients that minimizes the sum of the error between the value of each sample and the value obtained by multiplying the quantified value of the sample by a coefficient that corresponds to the position of the sample, of all the positions of samples, between several groups of predetermined coefficients that correspond to the positions of the samples, where each of the groups of coefficients is formed of coefficients arranged in a straight line in a plane that has frequency or time values corresponding to the sample positions with which the coefficients on a first axis are associated thereof and the values of the coefficients in a second axis thereof; and the coefficients of each of the various groups of coefficients are arranged in the plane in a straight line that has a different gradient from that of the straight lines for the other groups. characterized in that: the output of the number of bits of the index information at the stage of selecting coefficient groups is equal to or less than the value obtained by subtracting the number of bits actually used for a code corresponding to the vector quantization index of the number of bits assigned for the code corresponding to the vector quantization index.

Description

image 1

image2

image3

image4

image5

image6

5

10

fifteen

twenty

25

30

35

40

range of row numbers that can be identified by the idx index information that can be written in the region of unused bits as a search interval and select the row number m ’. In other words, the gradient calculator 116 selects only one row number m ’indicated by the idx index information that can be written in the region of unused bits. More specifically, the gradient calculator 116 selects only a row number m 'that can be identified by the idx index information that can be expressed with the number of bits not actually used for a code corresponding to the vector quantization index between the number of bits assigned for the code corresponding to the vector quantization index. For example, gradient calculator 116 identifies a row number m 'as given below, among the mMAX row numbers m = 0, ..., mMAx -1 that can be identified by the idx index information that can be written in the region of unused bits, and write idx index information that corresponds to the row number m 'in the region of unused bits.

image7

The symbol || · || indicates the rule of ·; argminm || || means that the minimization of || · || in m it becomes m '; argrninm means argrnin with subscript m; and  = [X (0),…, X (C0-1)],  ^ = [X ^ (0),…, X ^ (C0-1)]; and Am means a diagonal matrix that has gradient coefficient vectors m = [m (0), ..., m (C0 -1)] that correspond to the row number m as its diagonal elements, as shown then.

image8

The number of bits of the idx index information, described above, is equal to or less than the number of bits obtained by subtracting the number of bits actually used for a code corresponding to the vector quantization index of the number of bits allocated for the code that corresponds to the vector quantification index. From the above description, the idx index information can be transmitted simply using the region of unused bits.

[Example of step E4]

In this example, the gradient calculator 116 executes the steps shown in Fig. 3 to write the idx index information indicating the row number of the selected gradient coefficient vector in the region of unused bits. When C0 is L, the process of step E4 of Fig. 3 is executed for each frame. When C0 is a common divisor of L other than 1 or L, the process of step E4 of Fig. 3 is repeatedly executed for each sub-band in a single frame.

Gradient calculator 116 compares the input number of unused bits, U, with 0 (step E40); and if U> 0 is not satisfied, the gradient calculator 116 ends the process of step E4 without adapting the various quantized input values X ^ (0), ..., X ^ (C0 -1), as shown below .

image9

When U> 0 is satisfied, the gradient calculator 116 initializes m and idx by setting m = 0 and idx = 0 (step E41) and proceeds to step E42.

In step E42, the gradient calculator 116 uses the number of unused bits, U, to specify the range of row numbers that can be identified with the idx index information that can be written in the region of unused bits according to the search interval, and decides a mMAX search interval decision value to decide the search interval (the range of row numbers). In other words, gradient calculator 116 obtains mMAX by deciding the number of row numbers that can be identified with index information that can be written in the region of unused bits (step E42).

Normally, the number of unused bits, U, can identify 2u row numbers. Therefore, the search interval can be set to the range of 2 row numbers. In the present example, however, a value indicating that the correction with the use of the gradient coefficient vector m is not executed is assigned to

8

image10

image11

5

10

fifteen

twenty

25

30

35

40

Four. Five

image12

Gradient adjustment unit 124 reads the idx index information of the unused bit region of the modified vector quantization index, according to mMAX (step D33). The gradient adjustment unit 124 decides, for example, the position where the idx index information is stored, according to mMAX, and reads the idx index information.

Gradient adjustment unit 124 decides whether idx> 0 is satisfied (if idx = 0) (step D34). If Idx> 0 is satisfied (idx = 0 is not satisfied), the gradient adjustment unit 124 updates the various quantized values X ^ (b-C0), ..., X ^ ((b + 1) .C0 -1) , as shown below (step D35), and the process of step D3 ends.

image13

If idx> 0 is not satisfied (idx = 0 is satisfied), the gradient adjustment unit 124 ends the process of step D3 without updating the various decoded values X ^ (b.C0), ..., X ^ ((b +1) .C0 -1) (step D36), as shown below.

image14

The description of [Example of step D3] ends here.

If decoded signals in the time domain are necessary, the adjusted values X ^ UD (k) produced in the gradient adjustment unit 124, and the converter in the time domain 125 are entered into the time domain converter 125 transform X ^ UD (k) into signals in the time domain z (n), for example, by an inverse Fourier transform.

(Characteristics of this embodiment)

As described above, since in this embodiment, the decoder device 12 adjusts various decoded values X ^ (k) using the gradient coefficient vector selected by the encoder device 11, musical noise can be reduced and similar caused by the quantization error.

In this embodiment, a vector composed of gradient coefficients m (0), ..., m (C0 -1) that are correlated with each other is specified as a vector of gradient coefficients m. For example, the vector of gradient coefficients m is a vector composed of several gradient coefficients m (0), ..., m (C0-1) distributed asymmetrically, for example, in a straight line in the plane (k, m (k)). Input signals, such as audio signals or acoustic signals, often form a linear or curved envelope. By using the gradient coefficient vector m that reflects such characteristics of the input signals, the amount of idx index information can be suppressed while the quantization error is still adjusted with great precision. In the example shown in Fig. 6, the magnitude | X (k) | of the input signals in subbands k = 0, ..., 63 decreases as k increases. Therefore, when setting | X (0) |, ..., | X (63) | using the gradient coefficient vector m composed of gradient coefficients m (0), ..., m (63) distributed asymmetrically in a straight line with a negative gradient in the plane (k, m (k) ), your errors from the magnitudes | X ^ (0) |, ..., | X ^ (63) can be reduced | of the quantified values. By using the gradient coefficient vector m suitable for the characteristics of the input signals in each subband as described above, the quantization error can be efficiently reduced.

The idx index information to identify the gradient coefficient vector m selected by the encoder device 11, is transmitted using the region of unused bits efficiently, eliminating the need for an additional region to transmit the idx index information.

Modifications

The present invention is not limited to the embodiment described above. For example, if the decoder device 12 includes the flattening unit 126, the flattening unit 126 receives the set value X ^ UD (k) obtained in step D3 (Fig. 4) and, if a set value X ^ UD ( k) 'older than the adjusted value X ^ UD (k) is not 0, produces a weighted sum of the oldest adjusted value X ^ UD (k)' and the current adjusted value X ^ UD (k) as a flattened value X ^ POST (k). If X ^ UD (k) ’is 0, the flattening unit 126 does not obtain the weighted sum of the adjusted values, which means that the flattening unit 126 does not flatten the values

11 5

10

fifteen

twenty

25

30

35

40

Four. Five

adjusted, but produces X ^ UD (k) as X ^ POST (k) (step D4 in Fig. 4). Examples of the oldest adjusted values X ^ UD (k) 'include an adjusted value obtained in step D3 for the frame immediately preceding the frame corresponding to the adjusted value X ^ UD (k) and a flattened value obtained in the stage D4 for the frame immediately before the frame corresponding to the set value X ^ UD (k).

X ^ POST (k) is obtained by means of the following equations, in which α and β are adjustment factors and are determined appropriately depending on the requirements and specifications. For example, α = 0.85 and β =

0.15. α and β may vary appropriately depending on the requirements and specifications. ᶲ (ˑ) indicates a plus or minus sign of ˑ.

image15

Consequently, musical noise and the like caused by discontinuity over time in the amplitude characteristics of X ^ UD (k) can be reduced. If decoded signals in the time domain are necessary, X ^ POST (k) produced in the flattening unit 126 is introduced into the converter in the time domain 125. The converter in the time domain 125 transforms X ^ POST (k ) in signals in the time domain z (n), for example, by a Fourier Inverse transform.

It is not necessary that the input signals X (k) be signals in the frequency domain and can be any signals, such as signals in the time domain. The present invention can be applied to encode and decode any signal other than signals in the frequency domain. In this case, the gradient coefficients m (0), ..., m (C0 -1) that correspond to the same row number m are distributed asymmetrically in a straight line in the plane (k, m (k )), which has, for example, k (value that corresponds to the moment corresponding to the quantized value X ^ (k) to be multiplied by the gradient coefficient m (k), respectively, value that corresponds to the moment corresponding to the gradient coefficient m (k)) on its first axis and m (k) (gradient coefficient value) on its second axis. More specifically, the gradient coefficients m (0), ..., m (C0 -1) corresponding to the same row number m are placed, for example, in a straight line in the plane (k, m (k)). In this modification, k is a differentiated moment number that corresponds to a differentiated moment, and the positions of the samples X (k) are positions on the time axis that correspond to the numbers of differentiated moments k. If k is a differentiated moment number, a value greater than k corresponds to a later moment.

The step E3 can be executed so that in each frame a normalization value is determined FACTORY for the input signals X (k), the vector quantifier 115 uses a value obtained by normalizing the value X (k) of each sample of the signals Input with the FGANANCE normalization value instead of X (k) and use a value obtained by normalizing the quantified normalization value X- with the FGANANCE normalization value instead of X-. When step E3 is executed, X (k) can be substituted, for example, with X (k) / FGANANCE, and X- can be substituted with X- / FGANANCE. In that case, the normalization value calculator 112 is not necessary, and a value obtained by normalizing X (k) with the FGANANCE normalization value, instead of the quantized normalization value X can be entered into the normalization value quantifier 113 -. Next, the vector quantizer 115 can execute step E3 using a quantized value of a value obtained by normalizing X (k) with the FGANANCE normalization value instead of the quantized normalization value X-. The quantification index of the normalization value can correspond to a quantified value of a value obtained by normalization with the normalization value FGANANCE.

In the embodiment described above, the gradient calculator 116 of the coding device 11 decides whether idx> 0 is satisfied and, if idx> 0 is satisfied, updates several quantized values X ^ (b.C0), ..., X ^ ((b + 1) .C0 -1) or, if idx> 0 is not satisfied, it does not update the values (steps E410 to E412 in Fig. 3). The gradient adjustment unit 124 of the decoder device 12 decides whether idx> 0 is satisfied and, if idx> 0 is satisfied, updates several quantized values X ^ (b.C0), ..., X ^ ((b + 1). C0 -1) or, if idx> 0 is not satisfied, it does not update the values (steps D34 to D36 in Fig. 5). As a modification, a row vector (gradient coefficient vector) -1 = [1 (0), ..., -1 (C0-1)] = [1,…, 1] of the row number m = -1, composed only of elements "1" to the gradient matrix  given by Equation (1), and gradient calculator 116 and gradient adjustment unit 124 can calculate the following, independently if idx> 0 is satisfied.

12 5

image16

10

fifteen

twenty

25

30

35

40

Four. Five

Specific example values of row number m and idx index information do not limit the present invention. The numbers of m and idx given above may increase or decrease, and some of the numbers may not be used.

In the embodiment described above, the idx index information is stored in the region of unused bits of U unused bits, but the idx index information may not be stored in the region of unused bits.

The above-described processing can be executed in the order in which it is described or it can be executed in parallel or separately, depending on the performance of the apparatus that executes the processing or according to need. Other modifications may be made without departing from the scope of the invention.

Physical support, program and recording medium

The encoder device 11 and the decoder device 12 are configured by means of a known or general-purpose computer that includes, for example, a central processing unit (CPU) and a random access memory (RAM) and a special program in the that the processing described above be written. In that case, the special program is read by the CPU, and the CPU executes the special program to implement each function. The special program can be configured by a single program chain or can accomplish the objective by reading another program or library.

The program can be recorded on a computer readable recording medium. Examples of the computer readable recording medium include a magnetic recording apparatus, an optical disk, a magneto-optical recording medium and a semiconductor memory. Examples of the computer readable recording medium are non-transient recording media. The program is distributed, for example, by selling, transferring or lending a DVD, a CDROM or other transportable recording medium on which the program is recorded. The program can be stored in a memory of a server computer and can be distributed by transferring the program from the server computer to another computer through a network.

The computer that executes the program stores in its own memory the program recorded on a transportable recording medium or the program transferred from the server computer. When the processing is executed, the computer reads the program stored in its own memory and executes the processing according to the program read. The program can also be executed with other methods: The computer can read the program directly from the transportable recording medium and execute the processing according to the program; and each time the program is transferred from the server computer to the computer, the processing can be executed according to the transferred program.

At least a part of the process units of the encoder device 11 or the decoder device 12 may be configured by means of a special integrated circuit.

Description of reference numbers

11: Encoder device

111: Frequency domain converter

112: Normalization Value Calculator

113: Standardization Value Quantifier

115: Vector quantizer

116: Gradient Calculator

12: Decoder device

121: Decoder of normalization values

122: Vector decoder

124: Gradient adjustment unit

125: Time domain converter

126: Flattening unit

13

Claims (1)

image 1 image2